Bertini and Northcott (Q2658343)

The article under review shows that any suitably nice projective variety contains a curve whose height, genus and degree can be explicitly controlled. This is used to devise a general strategy which reduces the proof of estimates of Diophantine quantities associated to general abelian varieties to the case of Jacobians. Let us be more precise. Fix an algebraic closure \(\overline{\mathbb{Q}}\) of the field of rational numbers \(\mathbb{Q}\), and a ``sufficiently big'' infinite set \(S \subseteq \overline{\mathbb{Q}}\). Technically, this means that the \textit{Northcott number} \(m(S) := \inf\{ t \in \mathbb{R}_{\geq 0} \mid \# \{ s \in S \mid h_\infty(s) \leq t \} < +\infty \}\) is finite, where \(h_\infty \colon \overline{\mathbb{Q}} \to \mathbb{R}_{\geq 0}\) is the absolute logarithmic Weil height. The first main result of the paper under review (Theorem 1.3) shows that for every sub-field \(K \subseteq \overline{\mathbb{Q}}\), every integer \(N \geq 1\) and every geometrically irreducible, smooth, closed sub-variety \(X \subseteq \mathbb{P}^N_K\) of dimension \(\dim(X) \geq 2\), there exists a finite subset \(s \subseteq S\) and a non-singular, geometrically irreducible curve \(C\) defined over \(K(s)\) whose genus \(g(C)\), degree \(\deg(C)\) and height \(h_{\mathbb{P}^N_K}(C)\) can be explicitly controlled in terms of the degree \(\deg(X)\), the dimensions \(\dim(X)\) and \(N = \dim(\mathbb{P}^N_K)\), the Northcott number \(m(S)\) and the height \(h_{\mathbb{P}^N_K}(X)\). Here, \(h_{\mathbb{P}^N_K}\) denotes the height of a Chow form of a closed sub-scheme of \(\mathbb{P}^N_K\), as defined by \textit{P. Philippon} [Math. Ann. 289, No. 2, 255--283 (1991; Zbl 0726.14017)]. This can be viewed as a version of Bertini's theorem, with an explicit control of certain Diophantine quantities associated to the curve \(C\). The rest of the paper under review applies the aforementioned version of Bertini's theorem to abelian varieties \(A\). In this setting, it is more natural to consider the stable Faltings height \(h_F(A)\) instead of Philippon's height \(h_{\mathbb{P}^N_K}(\iota(A))\), which depends on the choice of an embedding \(\iota \colon A \hookrightarrow \mathbb{P}^N_K\). Lemma 2.2 of the paper under review provides a comparison between these two heights, using previous work of \textit{S. David} and \textit{P. Philippon} [Contemp. Math. 210, 333--364 (1998; Zbl 0899.11027)]. Using this inequality, Proposition 1.6 of the paper under review specialises Theorem 1.3 to show that for every principally polarized abelian variety \(A\) of dimension \(\dim(A) \geq 2\) which satisfies a technical condition and is defined over a number field \(K \subseteq \overline{\mathbb{Q}}\), there exists a finite subset \(s \subseteq S\), a curve \(C \subseteq A\) and a closed immersion \(A \hookrightarrow \mathrm{Jac}(C)\) defined over \(K(s)\), such that the genus of \(C\) is explicitly bounded in terms of \(\dim(A)\), and the stable Faltings height \(h_F(\mathrm{Jac}(C))\) of the Jacobian of \(C\) is bounded in terms of the dimension \(\dim(A)\), the Northcott number \(m(S)\) and the Faltings height \(h_F(A)\). The aforementioned technical condition, named \((\text{SS}\Theta)\) in the paper under review, assumes that the abelian variety \(A\) is semi-stable over \(K\), and that it is endowed with a suitable projective embedding. This can always be achieved up to passing to the division field \(K(A[48])\), as one can show combining results of \textit{D. Mumford} [Invent. Math. 1, 287--354 (1966; Zbl 0219.14024)] and \textit{M. Raynaud} [Sémin. Géom. Algébrique, Bois-Marie 1967--1969, SGA 7 I, Exp. No. 9, Lect. Notes Math. 288, 313--523 (1972; Zbl 0248.14006)]. Now, the authors provide a general strategy to reduce bounds on quantities associated to abelian varieties to the corresponding bounds for Jacobians. More precisely, fix a function \(q\) defined on pairs \((A,L)\) consisting of an abelian variety \(A\) defined over a number field \(K\), and of a finite extension \(L \supseteq K\). Assume, as explained in Definition 1.4 of the paper under review, that \(q\) ``behaves well'' with respect to products and isogenies of abelian varieties, and to the field extensions \(K \subseteq K(s)\) induced by finite subsets \(s \subseteq S\). Fix moreover an abelian variety \(A\) defined over \(K\), and suppose that there exists a curve \(C \subseteq A\) meeting the criteria of the curve constructed in Proposition 1.6. In particular, the curve \(C\) is defined over \(K(s_C)\) for some finite set \(s_C \subseteq S\). Then, the authors of the paper under review show in Theorem 1.5 that if one can bound \(q(\mathrm{Jac}(C),K(s_C))\) in terms of the dimension \(\dim(A)\) and the stable Faltings height \(h_F(\mathrm{Jac}(C))\), then one can obtain a similar bound for \(q(A,K)\). This result is applied in two different ways. First of all, the authors prove that the stable Faltings height \(h_F(A)\) of any abelian variety \(A\) defined over a number field \(K\) is bounded from below in terms of the dimension \(\dim(A)\), the degree \([K \colon \mathbb{Q}]\) and the norms of those primes of \(\mathcal{O}_K\) at which \(A\) has bad, semi-stable reduction. Second, the authors reduce Honda's conjecture on the rank of rational points to the case of Jacobians. More precisely, for any abelian variety \(A\) defined over a number field \(K\), \textit{T. Honda} has conjectured in [Jpn. J. Math. 30, 84--101 (1960; Zbl 0109.39602)] that there exists a real number \(c_A > 0\) such that \(\mathrm{rank}(A(M)) \leq c_A [M \colon \mathbb{Q}]\) for every finite extension \(K \subseteq M\). Then, Corollary 1.10 of the paper under review implies that if Honda's conjecture is true for all Jacobians \(J\) of curves defined over \(K\), and the real number \(c_J\) depends only on the dimension \(\dim(J)\) and the stable Faltings height \(h_F(J)\), then Honda's conjecture holds true for all abelian varieties \(A\) defined over \(K\), and \(c_A\) will depend only on the dimension \(\dim(A)\), the stable Faltings height \(h_F(A)\) and the Northcott number \(m(S)\). To conclude, let us mention that the kind of strategy provided in the paper under review, which reduces certain Diophantine properties from general abelian varieties to Jacobians, could come as a surprise, in view of the sparsity of Jacobians in the moduli space of abelian varieties. This is even true up to isogeny, as shown in the works of \textit{J. Tsimerman} [Ann. Math. (2) 176, No. 1, 637--650 (2012; Zbl 1250.14032)] and \textit{D. Masser} and \textit{U. Zannier} [Ann. Math. (2) 191, No. 2, 635--674 (2020; Zbl 1441.14101)].